SDM Baseline Condition Map Review
Seasonal Patterns in Sea Surface Temperatures
Seasonal Baseline Map Comparisons
Do baseline conditions look vastly different on a map? Are there Seasonal Patterns to the error?
Spring = March - May
Summer = June - August
August = September - November
Winter = December - February
Question that came up:
Should we scale the covariates and SSP scenarios based on their baseline mean/variance rather than the extracted data points mean/variance?
For that we would take a grand mean and variance for the year over our complete study area, and use that to scale with.
Would need mean/variance over the whole area and not by pixel to preserve the north and south patterns. But this would in theory avoid any of the spatial and temporal sampling biases.
Loading Baseline Period Means
2000 to 2019 In-Situ Average Conditions
2000 to 2019 SSP Bias-Corrected Baselines
Raw Values - Difference from Observed Baselines
Re-Scaled Value Maps
Before outputs from the SDM models can be applied to the projected landscapes, the environmental covariates need to be re-scaled to match how the model is fit. The following maps have been re-scaled using the grand mean and variance from the dataset used to fit the SDM model. These maps should still reflect the North/South spatial patterns of the raw data, but with units that have been centered and scaled.
@Adam Kemberling is going to work on maps of the fitted/projected covariates on the “scaled” scale. For those, the grand mean/SD for SST are: 11.1/4.5 and then BT are: 7/2.7. These will give us some idea of what the spatial patterns are with the “scaled” covariates. Ideally, we see patterns that are similar to the SST/BT patterns on the raw scale rather than just a somewhat random map with patches of positive/negative values.
These Are the Values the Model Uses to Project
